可以使用全局变量来作为另一个函数的参数吗？

huangapple
go评论101阅读模式
英文:

Can I use a global variable to parameter another function?

问题

我有一个bbox_sort函数，用于按从左到右的顺序排序坐标。为了考虑两个边界框之间的Y高度，我启动了thresh变量。在get_predict函数中，bbox_sort函数被作为参数传递给cmp_to_key。我需要使thresh成为可选参数。

我知道使用全局变量是不良实践，但我不知道如何将thresh参数传递给bbox_sort函数。将数值thresh写入一个单独的文件并从那里获取它是否是良好实践，或者我应该如何重新制作排序函数？

英文:

I have a bbox_sort function to sort the coordinates in order from left to right. To take into account the Y height between the two bounding bbox, I start the thresh variable. In get_predict , the bbox_sort function is passed as an argument to cmp_to_key. I need to make optional parameter thresh

I know it's bad practice to use global variables, but I don't know how else to pass the thresh parameter to the bbox_sort function. Is it good practice to write the number thresh to a separate file and get it from there or how can I remake the sort function?

  1. from functools import cmp_to_key
  3. def bbox_sort(a, b,):
  5.     bbox_a = a
  6.     bbox_b = b
  7.     
  8.     if abs(bbox_a[1] - bbox_b[1]) <= thresh: 
  9.         return bbox_a[0] - bbox_b[0]
  10.     
  11.     return bbox_a[1] - bbox_b[1]
  13. def get_prediction(result):
  14.     coord_list = []
  15.     res = result.to_coco_annotations()
  17.     global thresh
  18.     thresh = 10
  19.     for ann in res:
  21.         current_bbox = ann['bbox']
  22.         x = current_bbox[0]
  23.         y = current_bbox[1]
  24.         w = current_bbox[2]
  25.         h = current_bbox[3]
  26.     
  27.         coord_list.append((x, y, w, h))
  29.     cnts = sorted(coord_list, key=cmp_to_key(bbox_sort))
  30.     for pred in range(len(res)-1):
  31.         res[pred]['image_id'] = cnts.index(tuple(res[pred]['bbox']))
  33.     return res

答案1

得分: 1

  1. 你可以创建一个名为 `create_bbox_sort` 的函数，它会根据传递给 `create_bbox_sort` 的 `thresh` 参数动态创建一个 `bbox_sort` 函数。这样，你就不需要使用全局变量，而且 `create_bbox_sort` 函数可以轻松地在其他地方和其他脚本中使用。这类似于 [带属性的装饰器][1] 的工作方式。
  4. 新代码如下：
  6. ```python
  7. from functools import cmp_to_key
  9. def create_bbox_sort(thresh):
  11.     def bbox_sort(a, b,):
  13.         bbox_a = a
  14.         bbox_b = b
  15.         
  16.         if abs(bbox_a[1] - bbox_b[1]) <= thresh: 
  17.             return bbox_a[0] - bbox_b[0]
  18.         
  19.         return bbox_a[1] - bbox_b[1]
  20.     
  21.     return bbox_sort
  23. def get_prediction(result):
  24.     coord_list = []
  25.     res = result.to_coco_annotations()
  27.     thresh = 10
  28.     for ann in res:
  30.         current_bbox = ann['bbox']
  31.         x = current_bbox[0]
  32.         y = current_bbox[1]
  33.         w = current_bbox[2]
  34.         h = current_bbox[3]
  35.     
  36.         coord_list.append((x, y, w, h))
  38.     cnts = sorted(coord_list, key=cmp_to_key(create_bbox_sort(thresh)))
  39.     for pred in range(len(res)-1):
  40.         res[pred]['image_id'] = cnts.index(tuple(res[pred]['bbox']))
  42.     return res
  2. <details>
  3. <summary>英文:</summary>
  5. You can create a `create_bbox_sort` function which creates a `bbox_sort` function on the fly using the `thresh` passed as a parameter to `create_bbox_sort`. This way, you don&#39;t need a global variable and the `create_bbox_sort` function can easily be used elsewhere and in other scripts. This is similar to the way [decorators with attributes][1] work.
  8. The new code is :
  10. ```python
  11. from functools import cmp_to_key
  13. def create_bbox_sort(thresh):
  15.     def bbox_sort(a, b,):
  17.         bbox_a = a
  18.         bbox_b = b
  19.         
  20.         if abs(bbox_a[1] - bbox_b[1]) &lt;= thresh: 
  21.             return bbox_a[0] - bbox_b[0]
  22.         
  23.         return bbox_a[1] - bbox_b[1]
  24.     
  25.     return bbox_sort
  27. def get_prediction(result):
  28.     coord_list = []
  29.     res = result.to_coco_annotations()
  31.     thresh = 10
  32.     for ann in res:
  34.         current_bbox = ann[&#39;bbox&#39;]
  35.         x = current_bbox[0]
  36.         y = current_bbox[1]
  37.         w = current_bbox[2]
  38.         h = current_bbox[3]
  39.     
  40.         coord_list.append((x, y, w, h))
  42.     cnts = sorted(coord_list, key=cmp_to_key(create_bbox_sort(thresh)))
  43.     for pred in range(len(res)-1):
  44.         res[pred][&#39;image_id&#39;] = cnts.index(tuple(res[pred][&#39;bbox&#39;]))
  46.     return res

答案2

得分: 0

If your threshold is a global variable only configurable by the developer it's common practice to define it on top of the file written in all caps THRESHOLD = 10 right under your imports.

In python you only need to use global ... when you edit a global variable inside a function. (Which you don't plan on doing here, I assume.)

If the threshold isn't constant, and you want to be able to change it at runtime, you can wrap your function in a lambda like that:

cnts = sorted(coord_list, key=cmp_to_key(lambda a,b: bbox_sort(a, b, thresh)))

And then add a threshold to your arguments:

def bbox_sort(a, b, threshold):

英文:

If your threshold is a global variable only configurable by the developer it's common practice to define it on top of the file written in all caps THRESHOLD = 10 right under your imports.

In python you only need to use global ... when you edit a global variable inside a function. (Which you don't plan on doing here, I assume.)

If the threshold isn't constant, and you want to be able to change it at runtime, you can wrap your function in a lambda like that:

cnts = sorted(coord_list, key=cmp_to_key(lambda a,b: bbox_sort(a, b, thresh)))

And then add a threshold to your arguments:

def bbox_sort(a, b, threshold):

huangapple
  • 本文由 发表于 2023年2月23日 19:21:25
  • 转载请务必保留本文链接：https://go.coder-hub.com/75544136.html
  • function
  • global
  • oop
  • python
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定